{
  "id": 226510,
  "title": "A different mixup augmentation",
  "url": "/competitions/ranzcr-clip-catheter-line-classification/discussion/226510",
  "author_name": "Mariam Mohamed",
  "post_date": "2021-03-16T17:20:46.090000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I used an image size (2Lx2L), and put two images (with size LxL) side by side and fill the rest with zeros. This helps get my best single model on LB.<br>\nUsing big image size alone did not help me a lot but with this technique, it worked better.</p>",
  "messages": [
    {
      "id": 1240875,
      "postDate": "2021-03-16T17:20:46.090Z",
      "content": "<p>I used an image size (2Lx2L), and put two images (with size LxL) side by side and fill the rest with zeros. This helps get my best single model on LB.<br>\nUsing big image size alone did not help me a lot but with this technique, it worked better.</p>",
      "rawMarkdown": "I used an image size (2Lx2L), and put two images (with size LxL) side by side and fill the rest with zeros. This helps get my best single model on LB.\nUsing big image size alone did not help me a lot but with this technique, it worked better.",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1240875": "I used an image size (2Lx2L), and put two images (with size LxL) side by side and fill the rest with zeros. This helps get my best single model on LB.\nUsing big image size alone did not help me a lot but with this technique, it worked better."
  }
}